Noninvasive detection method and system for intelligent waiting chair

Abstract

The invention relates to the technical field of non-invasive detection, in particular to a non-invasive detection method and system for an intelligent waiting chair, and provides the following scheme that after a first signal is obtained, morphological opening and closing operation, interference removal and waveform optimization are conducted on the first signal, and after a second signal is obtained, a time axis is adjusted through a dynamic time warping (DTW) method, and signal stability is ensured. Finally, the heart rate, the oxyhemoglobin saturation and the blood pressure are analyzed and calculated through pulse wave characteristics, and accurate physiological parameter detection is provided; compared with the prior art, the method has the advantages that non-perception health monitoring is provided by using the intelligent waiting chair, the problem of signal instability caused by wearing tightness and gesture change of the wearable equipment is avoided, and meanwhile, the detection accuracy and robustness are improved by combining morphological filtering and DTW optimization signal processing; the method is suitable for hospitals, physical examination centers and other scenes, realizes efficient health monitoring in a waiting stage, and improves medical diagnosis and treatment efficiency.

Description

Technical Field

[0001] The present invention relates to the field of non-invasive detection technology, and in particular to a non-invasive detection method and system for an intelligent waiting chair. Background Art

[0002] Existing physiological signal detection technologies mainly rely on wearable devices (such as smart watches and bracelets) or manual measuring devices (such as medical blood pressure monitors and pulse monitors). These methods require users to actively wear or operate them during the detection process, resulting in inconvenience, low measurement frequency, and susceptibility to signal interference. In medical environments, traditional waiting area health monitoring mainly relies on manual measurement by medical staff. This lacks automation and continuity, resulting in the inability to grasp the patient's health status in real time while waiting, affecting the optimal allocation of medical resources. Although some smart waiting chairs have physiological signal detection functions, they usually rely on ECG (electrocardiogram) or simple PPG sensors, which are greatly affected by changes in user posture and small movements, resulting in unstable signal quality and limited detection accuracy.

[0003] The existing technology has the problems raised by this background technology: the detection method relies on the active cooperation of the user, the signal is easily disturbed by small movements, the detection accuracy is low and there is a lack of effective signal optimization processing, resulting in insufficient stability and automation level of health monitoring. In order to solve the above problems, this application designs a non-invasive detection method and system for smart waiting chairs. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to address the shortcomings of the existing technology and provide a non-invasive detection method and system for smart waiting chairs. The method obtains a first signal, performs morphological opening and closing operations on it, removes interference and optimizes the waveform. After obtaining the second signal, the dynamic time warping (DTW) method is used to adjust the time axis to ensure signal stability. Finally, the heart rate, blood oxygen saturation and blood pressure are calculated through pulse wave feature analysis to provide accurate physiological parameter detection. Compared with the existing technology, this method uses a smart waiting chair to provide non-perceptual health monitoring, avoiding the problem of signal instability caused by wearable devices due to wearing tightness and gesture changes. At the same time, it combines morphological filtering and DTW to optimize signal processing to improve detection accuracy and robustness. It is suitable for scenarios such as hospitals and physical examination centers, realizing efficient health monitoring during the waiting stage and improving the efficiency of medical diagnosis and treatment.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] The non-invasive detection method of the smart waiting chair is applied to the smart waiting chair, wherein the smart waiting chair is equipped with a PPG sensor fixed to the left and right armrests. The non-invasive detection method of the smart waiting chair includes:

[0007] collecting a first signal through the PPG sensor, wherein the first signal represents a pulse waveform of the user;

[0008] Performing morphological filtering on the first signal, performing opening and closing operations on the first signal using a structure element to remove peak interference and fill valleys, thereby obtaining a second signal;

[0009] A signal time axis of the second signal is adjusted, and a detection result is obtained according to the adjusted second signal.

[0010] Performing morphological filtering on the first signal includes:

[0011] Calculating a structure element according to the periodic characteristic of the first signal, wherein a length of the structure element is related to the periodic characteristic of the first signal;

[0012] performing an opening operation on the first signal, wherein the opening operation comprises performing an erosion operation on the first signal using the structure element;

[0013] performing a closing operation on the first signal, wherein the closing operation comprises performing a dilation operation on the first signal using the structure element;

[0014] A second signal is obtained, wherein the second signal is a signal obtained by processing the first signal through an opening and closing operation.

[0015] Calculating a structural element according to the periodic characteristics of the first signal includes:

[0016] calculating a periodic characteristic of the first signal;

[0017] Obtaining an initial structural element based on the period, peak, and trough information of the periodic feature;

[0018] The weight matrix of the initial structure element is adjusted by the least square method until the convergence condition of the loss function is met, and the structure element is calculated according to the final weight matrix.

[0019] The adjusting the weight matrix of the initial structure element by the least square method includes:

[0020] Setting an initial weight matrix and a penalty parameter according to the shape of the initial structure element;

[0021] Applying the initial structuring element to the first signal, performing filtering processing, and obtaining a filtered signal;

[0022] Calculating a difference between the filtered signal and the first signal to generate a loss function;

[0023] The loss value is calculated according to the loss function, and the weight matrix is iteratively adjusted by the gradient descent method.

[0024] The performing an opening operation on the first signal includes:

[0025] According to the length L of the structural element, the step size performing a window scan on the first signal;

[0026] In each sliding window, the corrosion operation is performed according to the local minimum value. If there is a signal point in the window with a fluctuation smaller than the local minimum value, it will be removed.

[0027] The performing a closing operation on the first signal includes:

[0028] Performing a window scan on the first signal with a step size B according to the length L of the structure element;

[0029] A dilation operation is performed in each sliding window according to the local maximum value to expand all signal points.

[0030] The adjusting the signal time axis of the second signal includes:

[0031] Obtain template pulse signal;

[0032] matching the second signal with the template pulse signal;

[0033] Calculate the DTW distance based on the matching result, and obtain the best matching path based on the DTW distance;

[0034] A signal time axis of the second signal is adjusted according to the best matching path.

[0035] The obtaining of a detection result according to the adjusted second signal includes:

[0036] Heart rate is calculated from the time intervals between successive pulse peaks;

[0037] calculating blood oxygen saturation according to a ratio of second signals of different wavelengths;

[0038] Calculate blood pressure using the pulse wave transit time model;

[0039] The heart rate, blood oxygen saturation and blood pressure are output as detection results.

[0040] A non-invasive detection system for an intelligent waiting chair, comprising a signal acquisition module, a signal processing module, and a result output module;

[0041] The signal acquisition module acquires a first signal through a PPG sensor, wherein the first signal represents a user's pulse waveform;

[0042] The signal processing module is configured to perform morphological filtering on the first signal to obtain a second signal, and adjust a signal time axis of the second signal;

[0043] The result output module obtains the detection result according to the adjusted second signal.

[0044] Signal processing module, including:

[0045] a filtering processing unit, performing an opening operation and a closing operation on the first signal using the structure element to remove narrow peak interference and fill narrow valleys;

[0046] The time alignment unit is used to adjust the signal time axis of the first signal after filtering.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] 1. This invention uses a fixed PPG sensor to achieve non-perceptual health monitoring, avoiding the signal instability problem caused by wearable device wearing tightness and gesture changes, and improving the convenience of detection and user experience;

[0049] 2. This invention uses morphological filtering to remove tiny motion artifacts and combines it with a dynamic time warping (DTW) algorithm to adjust the signal time axis, ensuring the stability and accuracy of physiological signals and effectively improving the detection accuracy of parameters such as heart rate, blood oxygen saturation, and blood pressure. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Other features, objects and advantages of the present invention will become more apparent upon reading the detailed description of non-limiting embodiments made with reference to the following drawings:

[0051] Figure 1 This is a flow chart of the non-invasive detection method of the smart waiting chair according to Example 1 of the present invention;

[0052] Figure 2 Schematic diagram of the placement of the PPG sensor in Example 1 of the present invention;

[0053] Figure 3 This is a schematic diagram of the morphological filtering process in Example 1 of the present invention;

[0054] Figure 4 This is a schematic diagram of the weight matrix adjustment process according to an embodiment of the present invention;

[0055] Figure 5 This is a module diagram of the non-invasive detection system of the smart waiting chair in Example 2 of the present invention. DETAILED DESCRIPTION

[0056] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0057] Example 1

[0058] See also Figure 1 The present invention provides an embodiment of a non-invasive detection method for an intelligent waiting chair, which is applied to the intelligent waiting chair. The intelligent waiting chair is equipped with a PPG sensor fixed to the left and right armrests. The specific steps of the non-invasive detection method are as follows:

[0059] S1: Collect the first signal through the PPG sensor:

[0060] In this embodiment, PPG sensors are fixed to the left and right armrests of the smart waiting chair to collect real-time photoplethysmography (PPG) signals from the user's fingers. The PPG sensors utilize multi-wavelength light sources (such as red and infrared) for detection, adapting to varying skin tones, subcutaneous tissue thickness, and ambient light interference, thereby improving signal acquisition stability. Furthermore, the sensors are fixed to the armrests, eliminating the need for the user to actively wear the device during signal acquisition. This avoids signal artifacts caused by variations in wear tightness or hand gestures associated with traditional wearable devices, improving the user experience and signal quality stability. Furthermore, to reduce ambient light interference, a light-shielding structure is provided around the sensors, resulting in a purer PPG signal and minimizing the impact of noise on subsequent data processing.

[0061] S2: Perform morphological filtering on the first signal to obtain a second signal;

[0062] In this embodiment, a signal processing method based on morphological filtering is employed to address PPG signal artifacts caused by subtle movements. First, the periodic characteristics of the first signal are calculated. Based on the pulse wave's peaks, troughs, and periodicity, the length and shape of the structural elements are dynamically adjusted to adapt to individual differences and pulse waveform variations under different physiological states. Next, the signal is corroded using a morphological opening operation to remove narrow peaks caused by slight hand tremors or changes in armrest pressure. Furthermore, a morphological closing operation is used to dilate the signal to fill in missing narrow troughs caused by subtle movement. This morphological filtering method effectively removes short-term abnormal signals while preserving the pulse wave's key features, improving signal smoothness and accuracy, and ultimately obtaining a high-quality second signal. Compared to traditional low-pass filtering, this step more accurately preserves the pulse wave morphology while removing artifacts, thereby enhancing the reliability of subsequent detection.

[0063] S3: adjusting the signal time axis of the second signal;

[0064] In this embodiment, slight hand movements during the measurement process may cause the PPG signal's time axis to shift, affecting the accuracy of physiological parameters calculated based on pulse wave features. To address this issue, a dynamic time warping (DTW) algorithm is used to adjust the time axis of the second signal. Specifically, a standard template pulse signal is first acquired. This template can be constructed based on historical data or a pre-set standard model. Then, the second signal is matched with the template signal, and the DTW distance between them is calculated to determine the optimal matching path. Finally, based on the optimal matching path, the time axis of the second signal is adjusted to align with the template signal, eliminating signal phase shifts caused by slight hand movements. This approach significantly improves the stability of the pulse wave features, ensuring more accurate calculations of key physiological parameters such as heart rate, blood oxygen saturation, and blood pressure. Compared to traditional fixed-window smoothing methods, the DTW method can dynamically adapt to individual user differences, improving the accuracy of time alignment and enhancing detection stability.

[0065] S4: Obtaining a detection result according to the adjusted second signal;

[0066] In this embodiment, key physiological parameters of the user are extracted from the second signal after morphological filtering and time axis adjustment. First, heart rate (HR) is calculated by detecting the time interval between consecutive pulse wave peaks. This method performs statistical analysis based on the pulse wave period and combines time and frequency domain analysis to improve detection accuracy. Second, blood oxygen saturation (SpO2) is calculated using the ratio of PPG signals from light sources of different wavelengths. The user's blood oxygen level is obtained by analyzing the AC / DC component ratio of red and infrared light and combining it with a calibration formula. Furthermore, to provide non-invasive blood pressure monitoring, this application estimates blood pressure based on the pulse transit time (PTT) model. Specifically, the time delay between the ECG R peak and the PPG peak is calculated, and combined with personalized calibration parameters, systolic and diastolic blood pressure are estimated. All detection results undergo signal quality assessment (such as signal-to-noise ratio calculation and autocorrelation analysis) to ensure the accuracy and reliability of the final output physiological parameters. Compared with traditional single-parameter detection methods, this method, based on morphological filtering, time alignment, and multi-feature fusion analysis, improves the accuracy of heart rate, blood oxygen, and blood pressure measurements, providing users with high-precision, non-invasive, and continuous health monitoring capabilities.

[0067] In current healthcare scenarios, health monitoring in waiting areas primarily relies on manual measurements or independent wearable devices such as blood pressure monitors, heart rate monitors, and oximeters. These methods present numerous challenges, including the requirement for active patient cooperation during the measurement process, the difficulty of managing decentralized equipment, and the inability to automatically integrate measurement data. This is particularly true in public healthcare settings such as hospitals and physical examination centers, where many patients are unable to obtain real-time information about their health status while waiting, and doctors are unable to assess patients' basic physiological indicators in advance, resulting in inefficient use of waiting time.

[0068] In this embodiment, the smart waiting chair, as an innovative medical auxiliary device, can provide a non-invasive health detection method that does not require additional wearable devices or additional operations, making the waiting process a non-perceptual, automated health data collection link. The smart waiting chair can complete the health data collection while the user sits down naturally, without affecting the user's normal waiting experience. Compared with traditional wearable devices or manual measurement methods, the application of smart waiting chairs can greatly improve the intelligence level of the waiting area, allowing patients to realize non-perceptual monitoring of their health data while waiting, and automatically transmit the data to medical staff or health management systems, so that doctors can grasp the patient's basic physiological information in advance before the consultation, thereby improving medical efficiency.

[0069] See also Figure 2 The embodiment of the present invention discloses a schematic diagram of the placement of a PPG sensor. Specifically, this embodiment adopts a non-invasive detection method based on PPG (photoplethysmography) on a smart waiting chair. Traditional PPG sensors usually move as the user wears them, and the signal is easily affected by the external environment and the user's own small movements during the acquisition process. For example, the natural shaking of the user's wrist, muscle contraction, blood vessel pressure fluctuations, etc. These factors will cause signal artifacts and lead to detection errors. The present application solution fixes the PPG sensor on the armrest of the smart waiting chair, allowing the user to perform the test in a resting state, avoiding the signal instability caused by changes in the wearing position, while also improving the user experience and making the detection process more convenient and comfortable. Especially in medical scenarios, such as hospital waiting areas, physical examination centers and other places, users can perform accurate non-invasive testing without wearing additional equipment, which improves the convenience and scalability of the test.

[0070] Furthermore, this embodiment processes the PPG signal using morphological filtering to remove artifacts caused by minor motion, and combines it with the dynamic time warping (DTW) method to adjust the signal's temporal axis to ensure the stability of detection results. For example, when a user's hand shakes slightly while sitting on a waiting chair, the PPG signal may experience transient fluctuations, causing pulse waveform distortion and affecting the subsequent calculation of physiological parameters such as heart rate, blood oxygen, and blood pressure. This embodiment uses morphological opening operations to remove narrow peaks caused by short-term motion artifacts, while closing operations fill in narrow valleys caused by minor displacements, resulting in a smoother signal that more closely matches the true pulse waveform. Furthermore, to address the timeline misalignment of the PPG signal caused by hand position deviation or contact pressure variations, this embodiment incorporates the DTW method, matching the real-time PPG signal with a standard pulse wave template to calculate the optimal time alignment path. This automatically corrects for time errors caused by individual differences or minor displacements, thereby ensuring the accuracy of physiological parameter calculations.

[0071] For example, suppose a patient is measuring his blood oxygen and heart rate in a hospital waiting area. Traditional wearable devices may cause fluctuations in measurement data due to different wearing tightness or unconscious small wrist movements of the patient, and even significant deviations in the test results at different time periods, affecting the doctor's judgment. However, the present application solution uses a fixed PPG sensor to ensure that the user is measured in a relatively static state, avoiding the degradation of signal quality due to hand shaking. At the same time, the morphological filtering method can effectively filter out short-term artifacts, making the pulse waveform clearer, and the DTW time alignment method ensures the continuity of the signal, making the final calculated physiological parameters more stable and reliable. Even if the user has a slight displacement during the test process, the time axis can be automatically adjusted to ensure the accuracy of the measurement data.

[0072] Preferably, in this embodiment, the non-invasive detection method provided by this application can also be applied to different types of smart waiting chairs to adapt to various medical scenarios and health monitoring needs, such as emergency waiting chairs, elderly care waiting chairs, physical examination center waiting chairs, etc., so that it can achieve high-precision, imperceptible health data collection and analysis in different application environments.

[0073] For example, in an emergency waiting chair, it can be used to quickly assess a patient's vital signs, helping medical staff prioritize patients with more serious conditions in a busy emergency environment. Since emergency patients may experience pain, anxiety, shortness of breath, and other physiological changes, traditional manual measurement methods are easily interfered with by external factors and are difficult to provide accurate data in the first place. The present application solution automatically collects heart rate, blood oxygen, pulse waveform and other data through a fixed PPG sensor, and combines morphological filtering to remove signal artifacts caused by slight hand shaking. At the same time, the DTW timeline is adjusted to optimize data quality, enabling emergency medical staff to obtain stable vital signs information before the patient enters the clinic, thereby improving the efficiency of emergency triage.

[0074] For example, in elderly care waiting chairs, the method can be used for long-term health monitoring, especially for elderly patients with chronic diseases, such as hypertension, cardiovascular disease, or diabetes. The physiological signals of the elderly are more susceptible to environmental influences than those of young people, and due to the decrease in vascular elasticity, traditional non-invasive blood pressure measurement methods are prone to errors under low blood flow conditions. This embodiment uses a fixed PPG sensor combined with a pulse wave transit time (PTT) model to dynamically adjust the signal processing parameters to adapt to the hemodynamic characteristics of the elderly and ensure the stability of blood pressure and blood oxygen measurements. At the same time, in the elderly care waiting area, users usually sit for a long time. This solution can use the waiting time to achieve uninterrupted health data collection, provide doctors with a more comprehensive health trend analysis, and improve the accuracy of chronic disease management.

[0075] In this embodiment, the PPG signal detection of the smart waiting chair faces a core challenge, namely how to effectively suppress signal artifacts caused by small movements and small displacements in a non-sensing measurement state while maintaining the main characteristics of the pulse wave. Compared with traditional wearable devices, the PPG sensor of the smart waiting chair of this application is fixed, and the user does not need to actively wear or adjust the device. Although this avoids the impact of the tightness and position changes of the wearable device, there are still problems such as slight hand movements, changes in armrest contact pressure, and unconscious slight shaking of the user. These factors can cause short-term fluctuations, narrow peaks, or signal amplitude drops in the PPG signal, which in turn affects the calculation of subsequent physiological parameters.

[0076] Specifically, while traditional signal processing methods, such as low-pass filtering, wavelet transforms, or adaptive noise suppression, are effective in certain application scenarios, they suffer from high computational overhead, complex parameter adjustments, and the potential for phase distortion. This is especially true in noninvasive testing, where the waveform characteristics of PPG signals are crucial for calculating physiological parameters. Any inappropriate smoothing or filtering operations can weaken the key information of the pulse wave and lead to measurement errors. Therefore, this embodiment employs a morphological filtering method, using opening and closing operations to adjust the local structure of the signal, effectively removing short-term artifacts while ensuring that the key morphology of the pulse wave is not affected.

[0077] Furthermore, morphological filtering processes the signal based on its morphological characteristics rather than frequency-domain filtering, making it particularly suitable for removing short, narrow peaks and troughs without changing the overall shape of the pulse wave. For example, when a user's finger shakes slightly, the PPG signal may experience a transient high-amplitude mutation (narrow peak), or a slight change in hand contact pressure may cause the PPG signal to briefly attenuate (narrow trough). Traditional filtering methods may smooth the entire signal, causing the waveform to become blunt or distorted. However, this application utilizes morphological opening operations to accurately remove short, narrow peaks, while simultaneously filling in narrow troughs through morphological closing operations, restoring the signal to a shape closer to the true pulse wave, thereby ensuring the stability and accuracy of the measurement results. In addition, morphological filtering is a windowed processing method that only involves local minimum and maximum value operations within the window. Its computational complexity is much lower than that of time-domain and frequency-domain filtering methods, making it particularly suitable for real-time, non-invasive detection applications in smart waiting chairs.

[0078] See also Figure 3 , a schematic diagram of the morphological filtering process according to an embodiment of the present invention, the specific steps of S2 are as follows:

[0079] S2.1: Calculate a structure element according to the periodic characteristic of the first signal, wherein the length of the structure element is related to the periodic characteristic of the first signal;

[0080] In this embodiment, the calculation of the structuring element is the most critical part of the morphological filtering process, as its length and shape directly affect the signal filtering effect. To ensure that the structuring element can adapt to individual differences between users and the fluctuating characteristics of the PPG signal, this embodiment dynamically adjusts the parameters of the structuring element by analyzing the periodic characteristics of the first signal.

[0081] Specifically, the pulse wave period is first extracted from the first signal. This period is the time interval between two consecutive pulse wave peaks, typically determined by calculating the time difference between local maxima (peaks). Furthermore, the pulse wave amplitude variations must be analyzed to ensure that the shape of the structuring element accurately matches the waveform characteristics of the signal. To this end, this embodiment employs a sliding window approach, traversing the data points of the first signal, screening for local maxima and minima, and calculating the average period of the pulse wave. Based on the pulse period, the length of the structuring element is set to approximately equal to or close to a certain proportion of the pulse wave period. This ensures that the morphological operation removes short-term interference signals without disrupting the overall waveform.

[0082] Furthermore, the morphology of the structuring element can adopt linear, elliptical, or exponential decay models to accommodate different signal waveforms. This embodiment prefers the use of dynamically adjusted linear structuring elements, which have the advantages of low computational complexity and good preservation of the primary morphology of the pulse wave. This ensures that the structuring element can both adapt to individual physiological characteristics and be adaptively adjusted to different signal morphologies, improving the accuracy and robustness of signal filtering.

[0083] S2.2: performing an opening operation on the first signal, wherein the opening operation includes performing an erosion operation on the first signal using the structure element to remove narrow peak interference caused by small motion;

[0084] In this embodiment, the purpose of the opening operation is to remove short-term artifacts caused by subtle hand movements, muscle contractions, or changes in contact pressure. These artifacts typically appear as abnormally narrow peaks and can interfere with the identification of normal pulse waveforms. To this end, this embodiment uses the structure element obtained from the aforementioned calculation to perform an erosion operation on the first signal.

[0085] Specifically, the first signal is scanned using a sliding window, with the length of each window matching the length of the structuring element. Within each window, the minimum value between the signal point and the structuring element is calculated, and the signal value within the window is eroded using this minimum value. This effectively eliminates interference from localized, narrow peaks. Since small motion artifacts often manifest as short, bursty signals that often last less than a complete pulse cycle, the erosion operation can weaken these brief, abnormal peaks, making the signal smoother.

[0086] Another key aspect of the preferred opening operation is ensuring that the normal pulse waveform is not excessively weakened. To this end, this embodiment introduces a dynamic weight adjustment strategy during the erosion process. During the erosion operation, the degree of erosion is adaptively adjusted based on the local fluctuation characteristics of the signal. This allows the structuring element to retain a certain proportion of the signal when processing high-amplitude peaks, ensuring that the main peak of the pulse wave is not completely lost. This method effectively removes motion artifacts while maximally preserving the true pulse information of the PPG signal, thereby improving the accuracy of subsequent calculations of physiological parameters such as heart rate and blood oxygen.

[0087] S2.3: performing a closing operation on the first signal, wherein the closing operation includes performing a dilation operation on the first signal using the structure element to fill narrow valleys generated by the small displacement;

[0088] In this embodiment, the purpose of the closing operation is to correct for missing narrow troughs caused by slight hand movements or changes in the signal acquisition environment. In non-static measurement environments, PPG signals may experience brief amplitude drops due to slight hand movement, uneven contact pressure, or hemodynamic changes. This can manifest as abnormally narrow or even broken troughs in the pulse waveform. To fill these abnormal troughs and make the signal more continuous, this embodiment uses the structure elements calculated above to perform an expansion operation on the first signal.

[0089] Specifically, within the sliding window, the local maximum value within the window is calculated, and all signal points within the window are expanded using the maximum value. This can effectively fill the local narrow valley area and make the pulse wave smoother.

[0090] To prevent loss of detail due to oversmoothing, this embodiment incorporates an adaptive smoothing algorithm during the expansion process. This algorithm adjusts the expansion amplitude of the structural elements upon detecting a signal break, compensating for signal loss without distorting the waveform structure. Compared to traditional low-pass filtering, the closing operation in this embodiment more effectively fills in signal gaps caused by minor motion, improving signal continuity and preserving the natural characteristics of the pulse waveform, making subsequent calculations of physiological parameters more accurate and stable.

[0091] S2.4: Acquire a second signal, where the second signal is a signal obtained by processing the first signal through an opening and closing operation;

[0092] In this embodiment, the second signal is a signal optimized by morphological filtering, and its waveform is smoother and more stable than that of the first signal. It can effectively reduce artifacts and short-term interference caused by small movements, thereby improving the availability of the signal.

[0093] Furthermore, in the process of acquiring the second signal, this embodiment can also evaluate the signal quality, including calculating parameters such as the mean square error (MSE) of the signal, the peak interval coefficient of variation (CV), and the signal smoothness, to ensure that the filtered signal meets the standard of high-quality data.

[0094] Preferably, to verify whether the second signal has residual artifacts, this embodiment introduces an abnormal signal detection mechanism. This mechanism uses autocorrelation analysis to determine the periodic stability of the signal and compares it with historical data. If significant residual artifacts are detected, the morphological filtering parameters are automatically adjusted, and steps S2.1 through S2.3 are re-executed to further optimize signal quality. This method ensures that the ultimately acquired second signal not only accurately reflects the user's physiological state but also possesses sufficient stability to support subsequent physiological parameter calculations and health status assessments.

[0095] It's important to note that in this embodiment, micro-motion and micro-displacement manifest differently in PPG signals, requiring different processing strategies. Micro-motion typically manifests as localized signal fluctuations, such as narrow peaks or valleys that appear over a short period of time. These fluctuations are primarily caused by slight hand tremors or involuntary muscle contractions, affecting the signal's transient stability. Therefore, morphological opening and closing operations are used to remove short-term artifacts and smooth the waveform. Micro-displacement, on the other hand, manifests as gradual changes in the overall signal amplitude or drift in the pulse wave timeline. These fluctuations are typically caused by slow hand movement or changes in contact pressure, affecting the signal's temporal consistency. Therefore, this embodiment uses dynamic time warping (DTW) to adjust the signal timeline and correct for phase offset. By combining these two processing methods, this embodiment effectively suppresses both short-term interference and long-term drift in PPG signals, ensuring signal continuity and stability, and providing more accurate data support for subsequent calculations of physiological parameters such as heart rate, blood oxygen levels, and blood pressure.

[0096] The specific steps of S2.1 are as follows:

[0097] S2.1.1: Calculate the periodic characteristics of the first signal;

[0098] In this embodiment, the first signal is a pulse waveform signal collected by a PPG sensor, and the calculation of periodic characteristics is a key step in obtaining basic structural information of the pulse waveform.

[0099] Specifically, a method combining zero-crossing detection and extreme point analysis is used to identify the main periodic components of the pulse wave, including peaks, troughs and their corresponding time intervals.

[0100] Preferably, to ensure accurate extraction of periodic features, this embodiment first preprocesses the first signal, including removing high-frequency noise and low-frequency baseline drift based on a bandpass filtering method to avoid interference components in the signal affecting the calculation of periodic features.

[0101] S2.1.2: Obtain initial structural elements based on the period, peak, and trough information of the periodic feature;

[0102] In this embodiment, the initial structuring element is derived based on the calculation results of periodic features. The primary purpose is to construct a morphological structuring element that is tailored to the characteristics of an individual's pulse waveform, thereby improving the denoising effect of morphological filtering. To ensure the compatibility of the structuring element with the pulse waveform, peak and trough information is first extracted for multiple complete cycles, and the average pulse waveform within these cycles is calculated to serve as a reference for constructing the structuring element. After obtaining the reference waveform, it is normalized to keep its amplitude distribution within an appropriate range, thereby improving the adaptability of the morphological filter across different individuals. Subsequently, the length of the structuring element is determined based on the temporal relationship between the peaks and troughs to match the pulse waveform period.

[0103] To enhance the stability of the structuring element during denoising, a morphological structure optimization strategy is preferably employed to modify the initial structuring element. This involves smoothing the morphology of the structuring element based on a multi-scale analysis method to better align it with the typical morphological characteristics of the pulse wave. This approach avoids structuring element mismatch caused by the large variability of individual pulse waves and improves the denoising accuracy and adaptability of subsequent morphological filtering.

[0104] S2.1.3: Adjust the weight matrix of the initial structure element by the least squares method until the convergence condition of the loss function is satisfied, and calculate the structure element according to the final weight matrix;

[0105] In this embodiment, to further optimize the morphology of the structuring element and improve the effectiveness of morphological filtering, the least squares method is used to adjust the weight matrix of the initial structuring element. The basic principle of this method is to introduce a weight matrix based on the initial structuring element to weight the structuring elements at different time points to improve its adaptability to pulse waveform characteristics.

[0106] Specifically, an objective function is first established, which aims to minimize the filtering error after the structuring element is applied to the PPG signal. The weight matrix is then adjusted through iterative optimization to minimize the difference between the filtered signal and the standard pulse waveform. During the iterative process, to avoid overfitting, a regularization strategy is used to constrain the weight matrix, ensuring that the adjusted structuring element retains its original morphological characteristics and does not over-adapt to certain special waveforms. Furthermore, to improve the convergence of the optimization process, a gradient descent method with an adaptive step size is used to make the optimization process more stable and avoid slow convergence or oscillation due to inappropriate step size selection. Finally, after the convergence conditions of the loss function are met, the optimized structuring element is determined and used for subsequent morphological filtering.

[0107] See also Figure 4 , a schematic diagram of the weight matrix adjustment process of an embodiment of the present invention. In this embodiment, the core purpose of the weight matrix adjustment is to optimize the structural elements used for morphological filtering so that it can more accurately adapt to the pulse waveform characteristics of different individuals, thereby removing motion artifacts while retaining the main physiological characteristic information of the pulse wave. Traditional morphological filtering methods usually use structural elements with fixed forms, such as rectangles, ellipses or custom templates, but these fixed-form structural elements cannot fully adapt to the differences in pulse waveforms between individuals, resulting in insufficient denoising capabilities or loss of key pulse waveform information in some cases. Therefore, this embodiment optimizes the weight matrix method so that the structural elements can be adaptively adjusted according to the local characteristics of the signal, thereby improving the effect of morphological filtering. The specific steps of weight matrix adjustment are as follows:

[0108] Setting an initial weight matrix and a penalty parameter according to the shape of the initial structure element;

[0109] Specifically, to ensure that the weight matrix converges to a reasonable optimal value during the adjustment process, an initial weight distribution needs to be set for it. This embodiment adopts a Gaussian weight distribution strategy, which assigns higher initial weights to the central region of the structuring element and lower weights to the edge regions. This ensures that during the optimization process, the core of the structuring element is more stable, while the edge regions can be dynamically adjusted based on the local characteristics of the signal. This weight distribution method helps enhance the smoothness of morphological filtering while preserving the main pulse waveform characteristics of the signal.

[0110] Furthermore, the initial weight matrix is set based on the basic form of the PPG signal, with higher weights assigned to pulse wave peaks and troughs, and lower weights set in transition regions (such as rising and falling edges). This ensures that key features are retained during filtering while avoiding excessive signal smoothing. Furthermore, to prevent extreme values in the weight matrix from affecting the optimization process, a penalty parameter is set. This penalty parameter constrains the range of variation of the weight matrix, preventing weight values from being too large or too small during the optimization process, thereby ensuring the stability and robustness of the filtering effect.

[0111] Applying the initial structuring element to the first signal, performing filtering processing, and obtaining a filtered signal;

[0112] Specifically, during the opening operation, the weight matrix influences the intensity of the erosion operation. Regions with higher weights result in stronger contraction of the signal, while regions with lower weights retain more signal features. This ensures that short-term interference caused by small movements is removed while preserving the main pulse wave morphology. During the closing operation, the weight matrix controls the extent of the expansion operation, ensuring that narrow troughs caused by small displacements are appropriately filled without over-expanding the waveform, thus ensuring signal integrity. During this process, a sliding window scanning method is used for filtering. The signal within each window is locally operated on the structuring element to obtain the optimized filtered signal.

[0113] The difference between the filtered signal and the first signal is calculated to generate a loss function, wherein the loss value of the loss function is calculated as follows:

[0114] ;

[0115] in, Represents the weight matrix The loss value, represents a single sampling point of the signal, Indicates the total number of sampling points of the signal, represents the value corresponding to the first signal of the i-th sampling point, Represents the value corresponding to the filtered signal of the i-th sampling point, represents the penalty parameter, represents a single element of the weight matrix, represents the total number of elements in the weight matrix, represents the jth element in the weight matrix;

[0116] Specifically, the loss function design considers the signal's time-domain characteristics. This not only focuses on the overall mean squared error (MSE) but also incorporates signal morphological characteristics. This ensures that the optimization objective is not only to minimize the error value but also to ensure that the filtered signal maintains waveform characteristics consistent with the original signal. Because the influence of subtle motion artifacts in PPG signals is complex, simple error calculations may not fully characterize signal quality. Therefore, this embodiment incorporates a signal structure constraint into the loss function, mitigating noise interference while preserving the original signal structure during the optimization process.

[0117] According to the loss value, the weight matrix is iteratively adjusted through the gradient descent method;

[0118] Specifically, the application of the gradient descent method enables the weight matrix to gradually converge to the optimal solution, making the final filtered signal more closely aligned with the true pulse waveform. During the iterative process, a dynamic learning rate is set: a larger learning rate is used in the early stages of optimization to accelerate convergence, while the learning rate is gradually reduced in the later stages of optimization to ensure the stability of the final weight matrix. Through multiple rounds of iterative calculations, the weight matrix is gradually adjusted to achieve the best morphological filtering effect, effectively removing motion artifacts while maximally retaining the key features of the signal, ultimately obtaining a higher-quality second signal.

[0119] The specific steps of S2.2 are as follows:

[0120] S2.2.1: Based on the length L of the structural element, the step size Perform window scanning on the first signal, where the step length is calculated as:

[0121] ;

[0122] in, represents the adjustment factor, To control the length Relative to the ratio of the pulse period T, ensure that the step size is not less than a minimum value related to the pulse period, Represents an adjustment parameter based on the signal change rate, ensuring that the step size is not too large when the signal changes rapidly, so as to accurately capture the rapidly changing characteristics. represents the sampling rate of the first signal, Indicates the instantaneous rate of change of the signal;

[0123] In this embodiment, to ensure that the morphological processing of the first signal is adaptive to the pulse waveform characteristics of different users, the step size of the window scan is dynamically adjusted rather than fixed. The step size affects the accuracy and computational efficiency of signal processing. If the step size is too large, local features may be lost, while if the step size is too small, the computational effort may increase and real-time performance may be reduced.

[0124] Specifically, the step size can be adaptively adjusted according to the signal characteristics, thereby balancing the computational efficiency and signal processing accuracy under different conditions. Determine the step size based on the morphological structure elements to ensure the effectiveness of the morphological operation; Combined pulse period T and sampling rate Calculate the step size so that it will not be less than a minimum value related to the pulse period, so as to avoid missing important signal features; Dynamic adjustment is performed based on the instantaneous rate of change V of the signal, reducing the step size when the signal changes dramatically to accurately capture rapidly changing waveforms, while increasing the step size when the signal is relatively stable to improve computational efficiency. This step size calculation method effectively adapts the signal characteristics of different users and different physiological states during morphological filtering of PPG signals, removing small motion artifacts while ensuring signal integrity, thereby improving the accuracy and stability of non-invasive testing.

[0125] S2.2.2: Perform an erosion operation based on the local minimum in each sliding window. If there is a signal point with a fluctuation smaller than the local minimum in the window, remove it.

[0126] In this embodiment, the purpose of the morphological erosion operation is to remove short-term, high-frequency noise caused by small movements, ensuring signal smoothness while avoiding the inadvertent deletion of important pulse wave features. Specifically, within the sliding window, the local minimum within the current window range is first calculated. This minimum value can be obtained through local neighborhood comparison, that is, the minimum signal value within the window is found and set as the erosion benchmark for the current window. Then, the signal points within the window are compared with this minimum value. If the amplitude fluctuation of a signal point is less than the minimum value, it is determined to be short-term noise or artifact and is removed.

[0127] As a preference, in practical applications, traditional corrosion methods may over-smooth the entire signal, resulting in the loss of some pulse wave features. To address this issue, this embodiment introduces an adaptive corrosion threshold. That is, when calculating the local minimum, the historical change trend of the signal is taken into account, and combined with a weighted average method, the signal points in the center area of the window are given a higher weight, while the weight of the boundary area is lower, thereby reducing the error caused by the boundary effect. In addition, this application also uses signal stationarity analysis. If the signal change in the current window is too large, the corrosion standard is appropriately relaxed to prevent the accidental deletion of important signals. When the signal change is small, the corrosion standard is tightened to improve the denoising ability.

[0128] The specific steps of S2.3 are as follows:

[0129] S2.3.1: Perform a window scan on the first signal with a step size B based on the length L of the structuring element;

[0130] In this embodiment, the window scan step size is determined similarly to that in S2.2.1 to ensure maximum coverage of critical signal regions during the scanning process while reducing redundant computation. The difference is that, in the dilation operation, the step size focuses more on the ability to fill the troughs. Therefore, in cases of rapidly varying signals, the step size may be appropriately enlarged to ensure that troughs do not experience signal loss. In cases of relatively stable signals, the step size may be reduced to ensure a precise filling process.

[0131] S2.3.2: Perform dilation operation based on the local maximum value in each sliding window to expand all signal points;

[0132] In this embodiment, the purpose of the dilation operation is to fill in missing troughs caused by minor displacements, thereby ensuring the integrity of the pulse waveform. Specifically, within the sliding window, a local maximum is first calculated, representing the maximum signal amplitude within the window. All signal points within the window are then moved closer to this maximum, thus filling in signal gaps caused by motion artifacts or changes in contact pressure.

[0133] To prevent excessive filling from affecting signal authenticity, this embodiment introduces a dynamic filling threshold control strategy. Under this strategy, when signal fluctuations within the window are small, the dilation operation applies a stronger filling force to ensure signal continuity. When signal fluctuations within the window are large, the filling force is appropriately reduced to avoid waveform distortion caused by excessive filling. Furthermore, when performing the dilation operation, this embodiment employs smooth interpolation technology. This means that the filling process not only simply moves signal points to local maxima but also interpolates data from adjacent windows, resulting in a more natural and smooth waveform after filling, avoiding sudden changes or over-adjustments.

[0134] The specific steps for S3 are as follows:

[0135] S3.1: Acquire template pulse signal;

[0136] In this embodiment, a template pulse signal is obtained to provide a reference signal for subsequent timeline alignment and signal matching. This template pulse signal can be based on two sources: a standardized pulse waveform derived from statistical analysis of a large dataset of PPG signals from healthy individuals; or a personalized template signal generated for an individual user using their historical measurement data.

[0137] Specifically, the acquisition of the template pulse signal requires consideration of signal stability and representativeness. Therefore, multiple cycles of PPG signals are averaged to reduce random noise in a single measurement. Furthermore, to ensure the template signal is suitable for different users, adaptive adjustments are made based on the user's pulse wave characteristics (such as waveform period, amplitude variation, rising and falling edge characteristics), ensuring a high degree of matching between different individuals.

[0138] Furthermore, when using historical data to generate personalized templates, signal quality assessment is performed to ensure that the selected historical data is free of artifacts and mutations. Morphological filtering is then used to further optimize the template signal to reduce the interference of short-term fluctuations on the matching process. Obtaining a high-quality template pulse signal improves the accuracy of subsequent matching, making timeline adjustments more precise, and thus enhancing the stability and reliability of physiological parameter calculations.

[0139] S3.2: Matching the second signal with the template pulse signal;

[0140] The purpose of this step is to find the best matching point between the signal to be measured (i.e., the second signal) and the template signal to ensure that key physiological features, such as peaks, troughs, rising edges, and falling edges, can be accurately aligned when the time axis is adjusted.

[0141] Specifically, the second signal needs to be preprocessed first to improve the accuracy of the matching. This includes normalizing the signal to eliminate the problem of waveform amplitude inconsistency caused by individual differences, and detrending the signal to eliminate the influence of low-frequency noise. Subsequently, the sliding window method is used to gradually scan the second signal, and the similarity between the signal in the window and the template signal is calculated to find the most likely matching area. During the matching process, the changes in the pulse wave cycle are also taken into account, and the length of the matching window is adaptively adjusted to ensure that signals under different heart rate states can still be accurately matched. In addition, a feature point weight mechanism is introduced in the matching process, that is, a higher weight is given to key points such as peaks and troughs, and the second highest weight is given to the changing trends of rising and falling edges, so as to ensure that the key features of the pulse waveform are focused on during matching.

[0142] S3.3: Calculate the DTW distance based on the matching result, and obtain the best matching path based on the DTW distance;

[0143] In this embodiment, dynamic time warping (DTW) is used to calculate the nonlinear time transformation relationship between the second signal and the template signal to find the optimal alignment path. The core concept of DTW is to allow for deformation along the time axis, ensuring that the second signal can still be aligned with the template signal even with slight displacements or rhythm changes. During the calculation, a distance matrix is first constructed between the two signals. Each element of this matrix represents the degree of difference between a point in the second signal and the corresponding point in the template signal. Then, a search is performed on this matrix to find the path with the lowest cumulative cost, i.e., the optimal matching path, which minimizes the overall error. During the path search, the system sets constraints based on the characteristics of the physiological signal, such as monotonicity constraints (to ensure that the matching path is monotonically increasing, avoiding time axis regression issues), continuity constraints (to ensure that the jumps between matching points do not exceed the physiologically possible range of variation), and local window constraints (to limit the search range to reduce computational effort and improve matching efficiency).

[0144] Furthermore, to further optimize matching results, a signal quality factor (SQI) is introduced into the DTW calculation process. If the matching quality of a signal segment falls below a set threshold, the weight distribution is adjusted, allowing the system to prioritize high-quality matching segments to improve overall matching results. This method ensures that even if the user moves slightly during the detection process or their physiological state fluctuates slightly, the signal can still be accurately matched through DTW, thereby achieving accurate adjustment of the timeline.

[0145] S3.4: Adjusting a signal time axis of the second signal according to the best matching path;

[0146] In this embodiment, based on the optimal matching path calculated by DTW, the time axis of the second signal is nonlinearly stretched or compressed to ensure its perfect alignment with the template signal. Specifically, the core of the time axis adjustment lies in mapping the key feature points of the second signal (such as peaks and troughs) to the corresponding positions of the template signal and interpolating or resampling the intermediate data points to smooth the transition. For example, if the matching path calculated by DTW shows that a section of the second signal is shorter than the template signal, additional data points are inserted in this area to stretch the time axis; conversely, if a section of the second signal is longer than the template signal, the data points in this area are merged or filtered to compress the time axis. In addition, to prevent signal distortion caused by time axis adjustment, the system adopts an adaptive interpolation strategy, that is, linear interpolation is used in areas with relatively gentle signal changes, and cubic spline interpolation is used in areas with more drastic waveform changes (such as peaks and troughs) to maintain signal smoothness and authenticity. To ensure the adjusted signal still conforms to physiological characteristics, a secondary verification is performed on the signal. This involves calculating the pulse wave period change rate before and after the adjustment and comparing it to a reasonable physiological range. If the adjusted signal period changes outside the normal range, the matching path is appropriately corrected to prevent signal anomalies caused by time axis adjustments. This method not only accurately corrects signal time axis deviations caused by slight user movement, but also ensures that the adjusted signal still conforms to physiological characteristics, improving the accuracy of subsequent physiological parameter calculations.

[0147] Obtaining a detection result according to the adjusted second signal includes:

[0148] Heart rate is calculated from the time intervals between successive pulse peaks;

[0149] calculating blood oxygen saturation according to a ratio of second signals of different wavelengths;

[0150] Calculate blood pressure using the pulse wave transit time model;

[0151] The heart rate, blood oxygen saturation and blood pressure are output as detection results.

[0152] Example 2

[0153] See also Figure 5 , the present invention provides an embodiment: a non-invasive detection system for an intelligent waiting chair, the system comprising a signal acquisition module, a signal processing module and a result output module;

[0154] The signal acquisition module acquires a first signal through a PPG sensor, wherein the first signal represents a user's pulse waveform;

[0155] The signal processing module is configured to perform morphological filtering on the first signal to obtain a second signal, and adjust a signal time axis of the second signal;

[0156] The result output module obtains the detection result according to the adjusted second signal.

[0157] Signal processing module, including:

[0158] a filtering processing unit, performing an opening operation and a closing operation on the first signal using the structure element to remove narrow peak interference and fill narrow valleys;

[0159] The time alignment unit is used to adjust the signal time axis of the first signal after filtering.

[0160] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A non-invasive detection method for an intelligent waiting chair, which is applied to an intelligent waiting chair equipped with PPG sensors fixed to the left and right armrests, is characterized in that: The non-invasive detection method of the intelligent waiting chair includes: collecting a first signal through the PPG sensor, wherein the first signal represents a pulse waveform of the user; Performing morphological filtering on the first signal, performing opening and closing operations on the first signal using a structure element to remove peak interference and fill valleys, thereby obtaining a second signal; A signal time axis of the second signal is adjusted, and a detection result is obtained according to the adjusted second signal.

2. The non-invasive detection method of the intelligent waiting chair according to claim 1 is characterized in that: Performing morphological filtering on the first signal includes: Calculating a structure element according to the periodic characteristic of the first signal, wherein a length of the structure element is related to the periodic characteristic of the first signal; performing an opening operation on the first signal, wherein the opening operation comprises performing an erosion operation on the first signal using the structure element; performing a closing operation on the first signal, wherein the closing operation comprises performing a dilation operation on the first signal using the structure element; A second signal is obtained, wherein the second signal is a signal obtained by processing the first signal through an opening and closing operation.

3. The non-invasive detection method of the intelligent waiting chair according to claim 2 is characterized in that: Calculating a structural element according to the periodic characteristics of the first signal includes: calculating a periodic characteristic of the first signal; Obtaining an initial structural element based on the period, peak, and trough information of the periodic feature; The weight matrix of the initial structure element is adjusted by the least square method until the convergence condition of the loss function is met, and the structure element is calculated according to the final weight matrix.

4. The non-invasive detection method of the intelligent waiting chair according to claim 3 is characterized in that: The adjusting the weight matrix of the initial structure element by the least square method includes: Setting an initial weight matrix and a penalty parameter according to the shape of the initial structure element; Applying the initial structuring element to the first signal, performing filtering processing, and obtaining a filtered signal; Calculating a difference between the filtered signal and the first signal to generate a loss function; The loss value is calculated according to the loss function, and the weight matrix is iteratively adjusted by the gradient descent method.

5. The non-invasive detection method of the intelligent waiting chair according to claim 2 is characterized in that: The performing an opening operation on the first signal includes: According to the length L of the structural element, the step size performing a window scan on the first signal; In each sliding window, the corrosion operation is performed according to the local minimum value. If there is a signal point in the window with a fluctuation smaller than the local minimum value, it will be removed.

6. The non-invasive detection method of the intelligent waiting chair according to claim 2 is characterized in that: The performing a closing operation on the first signal includes: Performing a window scan on the first signal with a step size B according to the length L of the structure element; A dilation operation is performed in each sliding window according to the local maximum value to expand all signal points.

7. The non-invasive detection method of the intelligent waiting chair according to claim 1 is characterized in that: The adjusting the signal time axis of the second signal includes: Obtain template pulse signal; matching the second signal with the template pulse signal; Calculate the DTW distance based on the matching result, and obtain the best matching path based on the DTW distance; A signal time axis of the second signal is adjusted according to the best matching path.

8. The non-invasive detection method of the intelligent waiting chair according to claim 7, characterized in that: The obtaining of a detection result according to the adjusted second signal includes: Heart rate is calculated from the time intervals between successive pulse peaks; calculating blood oxygen saturation according to a ratio of second signals of different wavelengths; Calculate blood pressure using the pulse wave transit time model; The heart rate, blood oxygen saturation and blood pressure are output as detection results.

9. A non-invasive detection system for an intelligent waiting chair, used to implement the non-invasive detection method for an intelligent waiting chair according to any one of claims 1 to 8, characterized in that: The system includes a signal acquisition module, a signal processing module and a result output module; The signal acquisition module acquires a first signal through a PPG sensor, wherein the first signal represents a user's pulse waveform; The signal processing module is configured to perform morphological filtering on the first signal to obtain a second signal, and adjust a signal time axis of the second signal; The result output module obtains the detection result according to the adjusted second signal.

10. The non-invasive detection system of the intelligent waiting chair according to claim 9, characterized in that: The signal processing module includes: a filtering processing unit, performing an opening operation and a closing operation on the first signal using the structure element to remove narrow peak interference and fill narrow valleys; The time alignment unit is used to adjust the signal time axis of the first signal after filtering.

Images